DE3317967C2 - Device for achieving heat transfer between a semiconductor wafer and a platen - Google Patents

Description

The invention relates to a device for achieving Heat transfer between a semiconductor die and one Platen according to the preamble of claim 1.

In the further processing of semiconductor wafers, for example for the manufacture of integrated circuits, it happens sometimes before exposing the semiconductor die to elevated temperatures will. These elevated temperatures are desirable for that Diffusion of impurities, epitaxial growth Layers, the application of high quality metal films or that Annealing of metallic semiconductor contacts and the like it is desirable to use thermal energy in a controlled and even manner to apply. In other applications, such as the Ion implantation and etching, thermal energy is one unwanted by-product. In these latter cases it may be undesirable to raise the wafers Let temperatures, for example, the additional Diffusion over prescribed limit values or the separation of Contamination at epitaxial interfaces are undesirable. In addition, at these elevated temperatures Photoresist layers are affected. This difficulty is used in the manufacture of highly integrated and very high integrated devices (LSI or VLSI) still tightened because a large number of processing steps in succession must be carried out. It is especially towards the end of Processing sequence a large number of doped areas, conductive layers or insulating layers on each intended location exists, and it is undesirable to do so disrupt physical characteristics by heat treatment. In these Cases, the wafers should be controlled, be cooled evenly. So it is desirable that Bring semiconductor chips to elevated temperatures when a  Process step this just requires, and they do the opposite cool to prevent it from reaching high temperatures when there is unwanted heat.

In previous approaches to achieve heat transfer from or to a semiconductor chip are on radiation, convection and conduction-based facilities have been in place. Semiconductors are made by infrared radiation exposed top has been heated, and one has Allow semiconductor chips to cool down with their own radiation. Semiconductors are made up of flows of heated gas elevated temperatures have been brought. Are semiconductor chips have also been inductively heated while on probes rest. Semiconductors have also been kept cool that the ion beam is guided intermittently and / or that Semiconductor wafer in the meantime on an actively cooled Metal plate has been moved, which limits throughput. The Metal plate was coated with grease or oil on which the Semiconductors were at rest or became electrostatic Force applied the semiconductor die against a slight compressible surface held on the actively cooled plate. For this, z. B. on the publication of L. D. Bollinger "Ion Milling for Semiconductor Production Processes ", Solid State Technology, November 1977. This known method and devices have been used in cooling semiconductor wafers not proven to be very effective when high ion fluxes or high Energy levels occur.

A convex platen on which a Semiconductor die is clamped, is known from US-A-4 282 924 which is the starting point in the formation of the generic term of Claim 1 forms. The cooling capacity of this device is through  the extent of the actual touch of the back of the Limited with the heat-conducting surface, because on microscopic level only small areas of the two surfaces (typically less than 5%) actually in with each other Come into contact.

In US-A-3 566 960 the difficulty becomes insufficient Contact between solid surfaces mentioned (see column 3, Line 2 ff.), And it becomes a circulating gaseous or liquid medium for cooling the workpiece in the vacuum chamber disclosed. The cooling of one lies on the same line Workpiece, preferably a semiconductor chip Gas heat conduction in a vacuum as described by M. King and P.H. rose in an essay "Experiments on Gas Cooling of Wafers ", in: Nuclear Instruments and Methods, Vol. 189, 1981, pp. 169 to 173 and in US-A-4,261,762. In this device, gas is in the middle of a cavity inserted behind a semiconductor chip. The gas turns one Heat coupling between a support plate and the Semiconductor chips achieved, as is typically the case with the  Technology of gas heat conduction takes place. In practice, however, it does to finite leakage losses due to imperfect seals, see that a pressure drop between the center of the cavity and the Scope exists. Because the thermal conductivity in a gas is proportional to the pressure, there will be more heat in the middle transmitted where there is higher pressure, and it results in a Temperature drop across the semiconductor wafer. With certain Processes such as metal coating do this Temperature drop due to uneven processing, the under Circumstances is undesirable. Since also the semiconductor chip is not pressed against a platen, it can feel like move a membrane freely as soon as there is any significant pressure in the gap inserted between the support plate and the semiconductor die becomes. Through the movement of the membrane Semiconductor wafer outwards, the gap is enlarged so that the heat transfer decreases, which is any gain in the Cancels transmission due to increased gas pressure.

The invention has for its object a device type specified at the beginning with improved mode of action create.

The task is achieved by combining the characteristics of the Claim 1 solved and by the further features of Subclaims designed and developed.

In the invention, the heat transfer from solid (Semiconductor plate) to solid (platen) with Gas heat pipe supported.  

To optimize the proportion of gas heat pipe, the gas pressure made as high as possible, but without the semiconductor chip too warp. This is achieved by clamping the semiconductor chip on a shaped plate with a convex surface.

With the invention, therefore, a heat transfer from or to one Semiconductors through gas support when touching solids reached. For this purpose, the semiconductor die is along a circumference pressed against the form or platen. Because of this A sufficiently strong system pressure is applied via the entire underside of the semiconductor chip is generated, so that the Gas pressure up to the size of this system pressure on the underside of the Semiconductor wafer can be applied without that Semiconductor chip from the platen. Gas under considerable pressure is placed in the microscopic cavities between the semiconductor die and the platen introduced while the gap remains almost unchanged. Since the Gap even at high gas pressures up to the level of the preload remains narrow due to the stretching, the thermal resistance reduced and the heat transfer promoted.

The invention is described below with further advantageous ones Details using schematically illustrated exemplary embodiments explained in more detail. In the drawings:

Fig. 1 is a representation from the aforementioned US-A-4 282 924, showing a device for heat transfer by solid state contact in semiconductor wafers;

Fig. 2 is an illustration from the above-mentioned US-A-4 261 762, showing a device for cooling due to gas heat conduction;

Fig. 3 shows a cross section through a device according to the invention;

FIG. 4 shows a top view of the device according to FIG. 3;

Fig. 5 is a graph of heat transfer in different gas pressure ranges;

6 is a graph of lifting a silicon wafer at the center thereof as a function of the gas pressure.

Fig. 7 is a graph of the heat transfer of nitrogen gas as a function of gas pressure in a typical semiconductor die that can stand out under the influence of gas pressure;

Fig. 8 is a Diagramin for explaining the heat transfer in the inventive apparatus as a function of the gas pressure;

Fig. 9 is a diagram of the overall heat transfer in a semiconductor wafer with a contribution by solid contact and a contribution by gas heat conduction.

The transfer of heat from one solid to another is a fundamental phenomenon in heat transfer. The terminology used depends on the particular use; but for Transfer always belongs to the transition in one way or another Direction, d. H. both heating and cooling, see e.g. B. H. Grober et al. "Fundamentals of Heat Transfer", Part 1 "Conduction of Heat in Solids ", 1961. Ideally, the two stand Surfaces of the respective touching solids without gaps in full contact. In practice, however, for example  one pressed against the surface of a platen Semiconductor chips, there are irregularities on the two Microscopic surfaces. Even if a Semiconductor plate is pressed firmly against the platen, is therefore the actual touch area in microscopic Scale significantly less than 10% of the total surface area in the field of semiconductor production. That makes it Heat transfer from solid to solid anything but that largest possible, especially for semiconductor wafers that typically processed in a vacuum where there is none Contributes to the heat transfer by convection or conduction. The factors affecting the effectiveness of heat transfer from Determine solids to solids in a vacuum are from M. G. Cooper et al. in "Thermal Contact Conductance", Int. J. Heat Mass Transfer, volume 12, page 279, 1969.

Figs. 1A to 1C show a semiconductor cooling device 10 disposed within an ion implantation chamber. In the unlocked position, the die cooling device 10 is in a position to receive a die 13 , as shown in FIG. 1, which is inserted into a slot 11 by gravity. The device 10 is then, as shown by an arrow, rotated about a bearing axis 10 into the vertical position according to FIG. 1B, specifically perpendicular to the path of an ion beam. During the positioning process, the plate is held in position by a clamping ring and bent under tension against a convex clamping plate. After implantation is complete, the die cooler, as the arrow shows, is rotated about the bearing axis 12 to the position shown in FIG. 1C to eject the die 13 either by gravity or by an ejector pin.

As an example of the technique of gas heat conduction used in semiconductor wafers, reference is once again made to the aforementioned publication by M. King and PH Rose and to US Pat. No. 4,261,762. As shown in FIG. 2, a semiconductor die 21 is arranged above a support plate 23 , with a gap between them in which a gas is introduced through a gas channel 22 with the aid of the devices 25 , 26 and 27 . Heat is transferred through the gas between the semiconductor die 21 and the support plate 23 . The pressure of the gas is necessarily below the pressure which would lift the semiconductor die 21 from the support plate 23 , which would nullify the basic purpose of the heat transfer. Even if the semiconductor die 21 is clamped firmly against the support plate 23 , the maximum permissible gas pressure is the pressure at which the semiconductor die 21 , which resembles a thin membrane, would essentially begin to deform away from the support plate 23 .

As shown in FIG. 6, the deformation of a 100 mm semiconductor die becomes considerable at a pressure of 1 Torr. Such a deformation would therefore be unacceptable, since the heat transfer would be greatly deteriorated if the gap between the support plate and the semiconductor plate increased. This can be seen from curve a in FIG. 7, where the heat transfer as a function of the pressure for the deformed semiconductor plate is shown in FIG. 6. When the semiconductor die begins to deform in the middle, the thermal conductivity for the semiconductor die is rapidly deteriorated. The heat transfer capacity of a pure gas heat conduction technology is therefore limited to that which can be achieved at pressures below about 2 torr.

In the embodiment of the device according to the invention, which is shown in cross section in Fig. 3 and in plan view in Fig. 4, gas is introduced through a line 44 into an annular recess or an annular channel 37 which circumscribes the upper surface of a mold plate 36 . Gas is introduced through the annular channel 37 around the circumference of the semiconductor wafer in the vicinity of the point at which the semiconductor wafer is clamped at the top and sealed from below (at 45 ). The contact pressure between the surface of the mold plate 36 and the back of a semiconductor die 41 is generated by the application of a clamping force on the part of a clamping device 42 . This pressure is chosen so that it is below the pressure at which the semiconductor chip 41 would break, but is strong enough to be able to take up a significant gas pressure behind the semiconductor chip. When gas is introduced under pressure from a source 43 through a valve 40 and line 44 into the annular channel 37 , it fills the microscopic cavities in the interface between the underside of the die 41 and the top of the mold plate 36 . When gas is introduced, part of the pressure that holds the die in a curved shape is now supplied by the gas. As the pressure increases, more of the force on the semiconductor die is contributed by the gas within the microscopic cavities, while less of the force is contributed by the solid state contact of the semiconductor die with the upper surface of the mold plate. Even with increased gas pressure, the semiconductor die remains in place until the gas pressure equals the preload pressure. Then the semiconductor chip is no longer fully supported by the mold plate, but stands out from the tips of the solid surface. The semiconductor wafer therefore begins to behave like a membrane which is subject to bending under the gas applied. Basically, the gas pressure balances the force exerted by the clamping device 42 , and the semiconductor chip floats above the surface of the mold plate. Any increase in pressure above this level is comparable to the application of the same increase in pressure to a semiconductor wafer that is not preloaded. As shown in FIG. 6, the semi-conductor plate begins to deform under the application of only 1 to 2 torr of excess pressure, and the thermal conductivity is greatly deteriorated. This is indicated by part c in Fig. 8, where the net heat transfer in the device according to the invention decreases rapidly when the pressure exceeds the contact pressure of the preload.

Ultimately, the determination of the heat transfer capacity of a platen 35 is based on the heat capacity of the heated or cooled fluid circulating in channels 38 . The heat mass of the platen 35 is sufficiently large that the platen 35 appears as a large heat source or sink for the semiconductor chip 41 (typically the semiconductor chip has a mass of approximately 4 g).

In the device and the method according to the invention the advantages of heat transfer from solid to solid with the support of heat transfer through gas conduction united. The support through gas heat pipe can be considerable be, since very significant gas pressures can be achieved.  

As shown in FIG. 8, at a bias pressure of 35 torr, the pressure behind the die can reach almost 35 torr without the die being lifted off. With the device according to the invention, solid-to-solid contact between the semiconductor plate and a clamping plate is achieved by preloading the semiconductor plate against the clamping plate in order to achieve a large, preferably uniform, contact pressure across the semiconductor plate. This is typically in the order of 30 to 50 torr, but can also be lower or higher. The upper limit for the preload is the pressure at which some semiconductor wafers would break. With external visual inspection, the semiconductor wafer would be seen tightly clamped to the platen. However, on a microscopic scale, the area in contact is still significantly less than 10% of the total surface area available. This is the case, regardless of whether the platen has a metallic or a resilient polymer surface. Gas is introduced into these microscopic cavities under pressure. The pressure can be increased to the level of the preload, since the pressure of the gas in the cavities replaces the lifting force provided by the tips on the surface of the platen. In essence, the gas pressure is increased while the gap between the backing plate and the semiconductor die remains almost unchanged, so that a significant gas pressure can be applied without the semiconductor die bending or lifting off.

In order to be able to fully assess the character of the invention (solid-to-solid body heat transfer in connection with gas heat pipe), it is useful to take a closer look at the mechanism of heat transfer through the gas heat pipe. As shown in Fig. 5, at low pressures, the rate of heat transfer by gas transfer increases linearly with pressure. Here, the density of the gas increases with increasing pressure, and the mean free path remains long enough so that the gas collisions predominantly occur either with the semiconductor plate or with the platen. The gas molecules essentially migrate back and forth between the semiconductor wafer and the platen. This pressure condition is called molecular flow. The higher the pressure in this area, the higher the heat transfer between the semiconductor wafer and the platen. For most of the gases of interest for semiconductor production, the molecular flow range is below approx. 1 torr. At sufficiently high pressures or sufficiently large gaps between the semiconductor plate and the platen, the gas collisions begin predominantly under gas molecules instead of with the semiconductor plate or with the platen. This is called the area of laminar flow. There is a transition area between the areas of molecular and laminar flow, in which properties of both areas are present. This range, in which some laminar flow characteristics are given, is above approximately 5 torr for most of the gases of interest in semiconductor production. In this area, the gas begins to behave at least partially like a fluid in which the thermal conductivity is independent of the pressure. As soon as this condition is reached, increasing the pressure at a given gap no longer brings any advantage. The heat resistance is reduced only by reducing the gap between the semiconductor die and the platen. The transition from molecular to laminar flow is gradual and takes place at different values, depending on the gap between the platen and the semiconductor wafer in the arrangement.

In the laminar flow area, the resistance to thermal conductivity is pressure-independent and gap-dependent. From this dependency is shown in Fig. 7. The laminar flow area or the transition area with components of laminar flow creates the horizontal curves where the heat transfer capacity is independent of the pressure. Here, the transition area is folded (see Fig. 5) in the pure laminar flow area. When the transition area or the area of pure laminar flow is reached, the heat transfer capacity becomes constant for a given distance between the semiconductor die and the platen. This ratio is generally recognized in the field of heat transfer, see e.g. BH Grober et al. "Fundamentals of Heat Transfer", Chapter 7, "Conduction in Rarefied Gases", p. 150 ff., 1961 and S. Dushman "Scientific Foundations of Vacuum Technique", 2nd edition, p. 43, 1962. The device according to the Invention works under this pressure condition where laminar flow is present. In order to be able to work under this condition, the semiconductor die is clamped along its circumference with a circumferential clamping force, as shown schematically in FIG. 3. This clamping is similar to the known clamping device 10 according to FIG. 1, but has an additional purpose. In the prior art, a semiconductor chip 13 is held by a circumferential clamping ring against a convex platen in order to achieve good solid-to-solid contact over the entire surface of the semiconductor chip. In the case of the inven tion, the contact from solid to solid is achieved and, in addition, an average narrow gap between the backing plate and the semiconductor plate is produced by the load, so that the optimal support is obtained by gas transmission. So there is not only the heat transfer component from solid to solid, but also the component by gas support. The combination of these two components is shown in Fig. 9, in which the curve a represents the transmission component due to the contact between the chipboard and the semiconductor wafer, while curve b represents the contribution of the gas heat pipe and curve c the net transmission. When the pressure is raised to the pre-load pressure, the solid state contact component is reduced until the die lifts or contact with the die is lost. The gas heat line component according to curve b increases with increasing pressure until the area of pure laminar flow is reached, and then remains essentially constant with the pressure.

In the device according to the invention, it has been found that contact pressures can easily be achieved by preloading 35 torr or more, so that appropriate gas pressures can be aimed at while maintaining the narrow gap. These pressures are readily high enough to offer something like laminar flow characteristics (see Figure 5). The load not only enables sufficiently high gas pressures but at the same time reduces the distance between the semiconductor platelet and the clamping plate to a minimum, as a result of which the heat transfer capacity of the gas within the microscopic cavities between the clamping plate and the semiconductor platelet is increased. Overall, the platen has a convex shape. Preferably, it has a smooth finely sleeked metal surface, e.g. B. made of soft aluminum. It has been shown that such a solid metal surface provides the preferred thermal contact from solid to solid, ie is more suitable than a resilient polymer coating. The quality of the thermal contact is directly proportional to the conductivity of the metal, inversely proportional to the hardness and proportional to the frequency of the peaks, which protrude beyond the roughness in the surface, see MG Cooper, see above. Flexible, thermally conductive polymers can be used; but overall they are not that high. In a preferred embodiment, the curvature of the platen is selected so that there is a uniform contact pressure across the semiconductor wafer when it is preloaded.

The operation of the device according to the invention in contrast to pure gas transmission for heat transfer according to the prior art can be seen from Fig. 8. As shown in FIG. 2, the known cooling device has Gaslei tung a finite gap 21 between a semiconductor platelet 20 and a supporting plate 23. Since there is no preload against a mold plate, the semiconductor chip begins to deform at some torr, which increases the gap. The thermal conductivity then drops early, as curve b shows. In contrast, the thermal conductivity for the device according to the invention still increases, as curve a shows, until the area of laminar flow is reached. If the preload pressure exceeds 35 torr, the semiconductor die begins to deform, and here, too, the thermal conductivity decreases abruptly, as curve c shows.

The performance achievable with the device according to the invention showed up when implanting a 3 inch silicon semi-conductor with a coating of photoresist material a 2 mA As + ion beam at 180 keV. The silicon half terplättchen was attached to the device according to the invention stuck. Air was under a pressure of less than 30 torr between the platen and the semiconductor die guided. A surface area of 51 cm² was implaned animals. The energy density that occurred was higher than 6 watts / cm². It was on the whole surface of the half lead platelet no deterioration of the photoresist material captured.

Claims (8)

1. Device for achieving heat transfer between a semiconductor wafer ( 41 ) and a platen ( 35 ) in a vacuum processing chamber with the following features:

• a) the platen ( 35 ) is provided with a convex surface for receiving the semiconductor wafer ( 41 ) to form a solid-state contact with the back of the semiconductor wafer;
• b) a clamping device (42) is provided for clamping the semiconductor wafer (41) at its periphery against the clamping plate (35) to generate a bias contact pressure between the clamping plate (35) and semiconductor chip (41),
  characterized by the following features:
• c) a gas supply device ( 37 , 40 , 43 , 44 ) is provided for the introduction of gas into microscopic cavities between the surface of the platen ( 35 ) and the semiconductor wafer ( 41 ), the gas pressure being on the one hand in such a region in which the Thermal conductivity is independent of the gas pressure, and on the other hand is lower than the voltage before the system pressure between the platen ( 35 ) and the semiconductor plate ( 41 ), so that the solid-state contact of the semiconductor plate ( 41 ) on the platen ( 35 ) is essentially maintained.

2. Device according to claim 1, characterized in that the convex surface of the platen ( 35 ) has a contour which generates a uniform contact pressure over the surface area of the semiconductor wafer ( 41 ). 3. Apparatus according to claim 1 or 2, characterized in that the clamping device ( 42 ) on the semiconductor wafer ( 41 ) applies a biasing force and thus generates such a system pressure that the gas pressure in the range of 0.0066 to 0.133 bar (5 to 100 Torr) can lie without the semiconductor wafer being lifted or bent. 4. Device according to one of claims 1 to 3, characterized in that the gas supply device ( 37 , 40 , 43 , 44 ) comprises a channel ( 44 ) within the clamping plate ( 35 ) which is connected to a gas source ( 43 ) on the outside and opens into an open, annular channel ( 37 ) in the surface of the platen ( 35 ) at a radius that is smaller than that of the clamped semiconductor wafer. 5. The device according to claim 4, characterized in that the surface of the platen ( 35 ) outside the open, annular channel ( 37 ) is embedded in a sealing device ( 45 ) which seals the gas against the vacuum processing chamber. 6. Device according to one of claims 1 to 5, characterized in that the clamping plate ( 35 ) has channels ( 38 ) in its interior for the circulation of a heat transfer medium. 7. Device according to one of claims 1 to 6, characterized in that the convex surface of the platen ( 35 ) is metallic and finely finished. 8. The device according to claim 7, characterized, that the Surface consists of aluminum.